Identifying teleconnections and multidecadal variability of East Asian surface temperature during the last millennium in CMIP5 simulations

We examine the relationships in models and reconstructions between the multidecadal variability of surface temperature in East Asia and two extratropical modes of variability: the Atlantic Multidecadal Oscillation (AMO) and the Pacific Decadal Oscillation (PDO). We analyse the spatial, temporal and spectral characteristics of the climate modes in the last millennium, historical and pre-industrial control simulations of seven Coupled Model Intercomparison Project phase 5 (CMIP5)/Paleoclimate Model Intercomparison Project phase 3 (PMIP3) global climate models (GCMs) to assess the relative influences of external forcing and unforced variability. These models produce PDO and AMO variability with realistic spatial patterns but widely varying spectral characteristics. AMO internal variability significantly influences East Asian temperature in five models (MPI, HadCM3, MRI, IPSL and CSIRO) but has a weak influence in the other two (BCC and CCSM4). In most models, external forcing greatly strengthens these statistical associations and hence the apparent teleconnection with the AMO. PDO internal variability strongly influences East Asian temperature in two out of seven models, but external forcing makes this apparent teleconnection much weaker. This indicates that the AMO–East Asian temperature relationship is partly driven by external forcing, whereas the PDO–temperature relationship is largely from internal variability within the climate system. Our findings suggest that external forcing confounds attempts to diagnose the teleconnections of internal multidecadal variability. Using AMO and PDO indices that represent internal variability more closely and minimising the influence of external forcing on East Asian temperature can partly ameliorate this confounding effect. Nevertheless, these approaches still yield differences between the forced and control simulations and they cannot always be applied to paleoclimate reconstructions. Thus, we recommend caution when interpreting teleconnections diagnosed from reconstructions that contain both forced and internal variations.